[postgis-devel] ST_MakeValid() behaviour

[Sandro Santilli]

On Mon, Apr 19, 2021 at 04:22:37PM -0400, Yuri Astrakhan wrote:
> I second Paul's comments - I would usually want "just fix the broken
> geometry in a <best+efficient> way possible"() method as part of the
> ST_MakeValid(). Moreover, in some cases it may be good to use
> ST_MakeValid() in a low-probability of an error scenario (most data is ok
> except some outliers) - forcing everything to be predictably valid.
> 
> I would only want to choose the fixing algorithm if I have the in-depth
> knowledge of the specific breakage, and know the process that keeps leading
> to that breakage. Without it, I am more likely to choose whichever the
> current default flavour of the month is that fixes things in the most
> "humanly expected" and "quick" way, and would be ok if broken geometries
> would result in slightly different output after an upgrade.  Hope there
> weren't too many "quotes" here :)

How about ST_MakeValidQuickAndDirt() ? :)

--strk;